Enhancing titres of therapeutic viral vectors using the transgene repression in vector production (TRiP) system

A key challenge in the field of therapeutic viral vector/vaccine manufacturing is maximizing production. For most vector platforms, the ‘benchmark' vector titres are achieved with inert reporter genes. However, expression of therapeutic transgenes can often adversely affect vector titres due to biological effects on cell metabolism and/or on the vector virion itself. Here, we exemplify the novel ‘Transgene Repression In vector Production' (TRiP) system for the production of both RNA- and DNA-based viral vectors. The TRiP system utilizes a translational block of one or more transgenes by employing the bacterial tryptophan RNA-binding attenuation protein (TRAP), which binds its target RNA sequence close to the transgene initiation codon. We report enhancement of titres of lentiviral vectors expressing Cyclo-oxygenase-2 by 600-fold, and adenoviral vectors expressing the pro-apoptotic gene Bax by >150,000-fold. The TRiP system is transgene-independent and will be a particularly useful platform in the clinical development of viral vectors expressing problematic transgenes.

Supplementary figure 3. Assessment of therapeutic transgene protein expression on vector titres by vector genome mixing. The impact of the transgene protein on vector titres during production can be assessed by a simple mixing experiment. A reporter-encoding vector genome such as lacZ is mixed with therapeutic vector genome at the stated mass ratios of plasmid at transfection. The impact of transgene protein expression during vector production can be measured by reporter (lacZ) vector titration. Note that whilst GFP is considered to be inert, the effect of mixing a large proportion of GFP vector genome relative to lacZ (5:1) has the impact of reducing lacZ vector titres merely because proportionally more GFP-encoding vector particles are produced due to greater abundance of this vector genome RNA in the cell (a non-packageable therapeutic transgene expression plasmid can be used instead to minimise this effect). Therefore, data should only compared to GFP (no fill; negative control) at the same relative vector genome ratios.  were used to produce mixed EIAV-based vectors. Under 'standard' vector production protocol conditions, control vector genomes encoding either GFP or human Factor VIII were mixed 1:1 (mass ratio) prior to cotransfection of HEK293T cells together with EIAV vector packaging components, and either TRAP plasmid (+TRAP) or pBluescript (-). In the TRiP system, tbs-containing vector genomes encoding either GFP or Factor VIII were mixed 1:1 (mass ratio) prior to co-transfection of cells with packaging components. (a) Two days post-transfection, replicate cultures were analysed by flow cytometry to generate GFP Expression Scores for each condition i.e. to measure the direct effect of TRAP on GFP expression, and indirectly as a model for Factor VIII expression (note we were unable to develop a robust assay for human Factor VIII detection). Data are mean average values ± s.d. [log 10 -transformed data] (n=4); *p<8.4x10 (c) 50-fold concentrated vector preparations made in the stated cell lines were analysed by immunoblot for VSVG content (loading normalised by F-PERT assay); this is an indirect measure of Factor VIII activity as Factor VIII has been shown to inhibit VSVG incorporation into EIAV-based vector virions 1 . We have previously shown that subtle changes in VSVG pseudotyping can affect virion activity 2 , and this observation supports the notion that pre-existing pools of TRAP protein can further benefit vector production in the TRiPLenti system, particularly when the impact of transgene protein is potent. identified in duplicate EIAV-GFP vector preps to assess data quality. Only proteins that varied less than 4-fold between duplicate samples are included, and proteins of note are labelled with percentage proportion, which was generated by dividing the average Spectral index scores for each hit by the sum of all scores in the top 154 hits. EIAV Gag and Pol peptides appeared were detected at expected ratios, and VSVG/GFP were present in high abundance. Selected cellular proteins known to be incorporated into HIV-1 virions are highlighted. (b) Concentrated EIAV vector preps used for SINQ analysis were analysed by F-PERT (RT activity) and titrated against a standard EIAV-GFP prep of known biological activity to yield an arbitrary evaluation of titre (F-PERT-predicted titre), and also analysed by SDS-PAGE/Immunoblotting to EIAV capsid (p26) and VSV-G. Data are mean average values ± s.d. [log 10 -transformed data] (n=4); *p<2.9x10 -3 , **p<7.5x10 4 TUs of concentrated EIAV-tbsCOX2 into the front anterior eye chamber and assessed for inflammatory signals (this route of administration is extremely sensitive to inflammation), as well as immunological response to vector constituents. This was achieved by using sera from Day 0 or Day 44 post-injection as primary antibody to immunoblots of non-denaturing SDS-PAGE separated HEK293T cell lysates expressing individual vector components. (a) Positive control immunoblots indicating specific bands to VSV-G, EIAV Capsid (p26), COX2 (blot lane re-positioned relative to UTC for clarity) and TRAP. Note that under non-denaturing conditions the intact TRAP 11-mer is detected. (c-d) Tests performed on sera from three animals, indicating no immune response to vector components including TRAP[H6], except in one animal (4003) where antibodies to VSV-G were present at Day 44. All blots were exposed for the same amount of time. Vector was well-tolerated in this study with no vector-related inflammation and most ocular changes were secondary to the intracameral dosing procedure alone (data not shown). There were no treatment-related clinical signs or ophthalmic findings following dosing in any animal with vector in this study (data not shown). This is encouraging data that indicates the presence of TRAP in vector preps does not increase vector immunogenicity. . (c) Production of scAAV2-CMV-tbsGFP in HEK293T cells was repeated, together with scAAV2-CMV-tbsBax, which encodes the proapoptotic gene Bax. vDNA in crude vector material was quantified by qPCR using a primer/probe set to CMVp (only vector genome plasmid contained a CMV promoter). A negative control transfection was included (No RepCap) to control for residual pDNA within crude harvest material, which was substantial despite our efforts to reduce DNA using Benzonase ® . Nevertheless, we were able to detect scAAV-CMV-tbsBax vDNA at similar levels to scAAV-CMV-tbsGFP, only when TRAP plasmid was supplied during transfection. Data are mean average values ± s.d. [log 10 -transformed data] (n=3); *p<6.0x10  Figure 7d of the main report (see main report experiment details). Western blots were cut at two positions (between 19kDa and 26kDa, and between 26kDa and 37kDa markers; see arrows) and probed for GFP (~27kDa), Bax (~21kDa), cleaved Poly (ADP-ribose) polymerase (clv-PARP; ~89kDa), and GAPDH (~37kDa). The separate blots were re-aligned during ECL/film development.

Evaluation of the methods of GFP protein expression in transfected HEK293T cells
HEK293T cells were transfected with pHIV-CMV-GFP or pHIV-CMV-tbsGFP vector genome plasmids +/-pEF1a-coTRAP[H6]. Expression of GFP was carried out by flow cytometry (FACSVerse, BD Biosciences), outgating dead cells prior to analysis of FL1 channel events. GFP Expression scores were calculated by multiplying % GFP-positive cells by the median fluorescence intensity of events within the % GFP-positive gate. GFP expression in cell lysates was carried out by SDS-PAGE under reducing conditions, Western transfer and immunoblotting using antibodies to GFP (Ab290, AbCam). Species-specific HRP conjugated secondary antibodies were used at 1:1000 dilution.
Testing the level of IRES-dependent transgene expression from HIV-1-based vector genome plasmid DNA in HEK293T cells HEK293T cells were transfected with pHIV-Luc-IRES-GFP based vector genome plasmids +/-pEF1a-coTRAP[H6]. Vector genomes either contained an internal CMV promoter or no promoter, and were differentially controlled by TRAP/tbs at the ORF2 positions, encoding GFP. Expression of GFP was carried out by flow cytometry (FACSVerse, BD Biosciences), out-gating dead cells prior to analysis of FL1 channel events. GFP Expression scores were calculated by multiplying % GFP-positive cells by the median fluorescence intensity of events within the % GFP-positive gate.

Evaluation of impact of transgene by vector genome plasmid mixing experiments
The impact of therapeutic transgene encoded protein on vector titres was carried out by a vector genome plasmid mixing experiment. Therapeutic EIAV vector genome plasmid was mixed at either 1:1 or 5:1 mass ratio with an EIAV-lacZ vector genome plasmid during transfection of vector components. EIAV-GFP was used as a negative control using the same genome plasmid ratios so that lacZ genome dilution could be accounted for. Crude vector was generated as stated in Materials and Methods, and titrated by serial 10-fold dilution followed by transduction of D17 cells in the presence of 8µg per mL polybrene for 5-6 hours. Fresh media was added before a 3 day incubation, followed by fixation in formalin and staining with 0.04% X-Gal. Transducing events (2 or more clusters of blue cells) were counted, from which lacZ TU per mL titres were calculated. The lacZ titres generated for each mix with therapeutic vector genome must be compared to the equivalent GFP|lacZ control mix in order to evaluate the impact of transgene on vector titres. To assess the benefits of the stable TRiP system over the transient TRiP system, EIAV vector genome plasmids encoding human Factor VIII or GFP were mixed during production. HEK293T cells and HEK293T.TRiP cells (stably expressing TRAP[H6]) were used to produce mixed EIAV-based vectors. Under 'standard' vector production protocol conditions, control vector genomes encoding either GFP or human Factor VIII were mixed 1:1 (mass ratio) prior to transfection of HEK293T cells together with EIAV vector packaging components, and either TRAP plasmid (+TRAP) or pBluescript (-). In the TRiP system, tbs-containing vector genomes encoding either GFP or human Factor VIII were mixed 1:1 (mass ratio) prior to transfection of HEK293T.TRiP cells together with EIAV vector packaging components and either TRAP plasmid (+TRAP) or pBluescript (-). Two days post-transfection, replicate cultures were analysed by flow cytometry to generate GFP Expression scores for each condition i.e. to measure the direct effect of TRAP on GFP expression, and indirectly as a model for human Factor VIII expression (note we were unable to develop a robust assay for human Factor VIII detection). Crude vector harvests generated were titrated by GFP transduction assay in HEK293T cells to measure impact of human Factor VIII expression on vector titres. 50-fold concentrated vector virions were analysed by immunoblot for VSVG content; an indirect measure of human Factor VIII activity (human Factor VIII has been shown to inhibit VSVG incorporation into EIAV-based vector virions 14 ).

Non-SINQ analysis of EIAV-GFP, EIAV-COX2 and EIAV-tbsCOX2 vector preparations
Physical titration of concentrated EIAV vector particles was carried out by F-PERT 3 and SDS-PAGE was carried out on samples by loading either equal volumes or by F-PERT-normalised volumes. Immunoblotting was carried out using anti-VSV-G (AFPA1-30138, Fisher) and anti-EIAV p26 (capsid; mAb 12E8.1) antibodies. Species-specific HRP conjugated secondary antibodies were used at 1:1000 dilution.
Testing putative antibody response to EIAV-tbsCOX2 vectors produced using the TRiPLenti system in vivo by rat serum analysis A volume containing approximately 10 4 transducing units (HEK293T DNA integration assay units) of ~2000fold concentrated (double centrifugation) EIAV-tbsCOX2 vector was delivered into the front of the anterior chamber (intracamerally) of Wistar Hannover rat eyes (Crl:WI(Han); male, ~7 weeks old at time of dosing). This was part of a 56 day study to investigate gene transfer as part of the development of a gene therapy for primary open-angle glaucoma. Serum samples at day 0 and 44 were analysed by use as primary antibody samples (1:200 dilution) against Western blots. Western blots were generated by non-denaturing SDS-PAGE